The statement you referenced is spoken from someone who doesn't understand what they are talking about. There were two suggestions given. I gave one; you can try the other above, as evidenced by his post, not all computers rely on alternate power outside bios battery for clock time, during PC on state.
Thanks, that's not what I meant. The orginal PC RTC was a Motorola part with an embedded oscillator. New parts use an external reference, which is usually from the CPU's clock, divided as necessary.
And that CPU clock is a quartz crystal referenced oscillator. I'd like to see that embedded oscillator chip you referred to that didn't use a quartz crystal as a reference. Anyways, you probably know more than most here, so sorry to rib you on that... Not possible to snowball me in this subject area, however.
Right, I meant an internal crystal. Here's a reference to the part's spec sheet: http://memnon.ii.uam.es/~eloy/presentaciones/MC146818.pdf
wrong again. See the figure 1 block that says time base input/Osc, pins 2 & 3 (OSC1 AND OSC2). That is referring to an EXTERNAL crystal oscillator. Sorry, it is not embedded in the chip. Figs. 10&11. also makes the distinction more explicit, showing how to connect the external crystal or time base (which is ultimately referenced to a crystal as well). Let's kill this conversation, as it's a waste of time. You obviously know a bit about circuits though, are you a student in engineering?
Been a while, for sure... Not worth the time to dig, but one of the chips had an internal crystal while the other didn't, thought it was in the original and not the replacement, must be the other way 'round. Student? No. Old time Navy missle & search radar technician. Moved over to programming after I got out in '82.
Could be a flaky RTC chip ..... on top of that, Windows can't keep time well with ACPI turned on (not a problem for Linux), and the problem also gets even worse if you have dual core or quad core CPU. But don't worry about any of that, install the NTP service to transparently and optimally synch you up within milliseconds. That will fix it unless something is totally munged. http://www.meinberg.de/english/sw/ntp.htm
For all practical purposes all clocks drift. Reading the clock, by whatever method, can contribute to the drift. I have heard, though I do not recall the source, that the method employed by Windows to get, calculate and display the current time can create additional drift and the magnitude of that drift can be dependent upon machine load. I have observed that new machines running Windows can have a drift of about a minute within 1 - 2 days. One solution is to have a program run regularly and frequently (and unobtrusively) on your machine that synchronizes your machine's clock to a well known time source. I use Symmetricom. I am not sure, but i think it was free. I have it synching every 2 minutes and once it has synched the first time, subsequent changes it makes to the clock always seem to be no more than a few milliseconds.